Light-O1: a robot brain pretrained on human videos controls multiple humanoids
Scobleizer · x · 2026-09-21
LightOrigins introduces Light-O1, a whole-body intelligence model pretrained on structured human actions recovered from internet videos.
Key points:
- Human action pretraining at scale: leverages human action data extracted from internet videos, matching a scale and diversity that robot data collection alone struggles to reach.
- Cross-embodiment scaling law: scaling human action pretraining yields power-law reductions in prediction error across different embodiments.
- Whole-body intelligence: adapted to different humanoids and hands, Light-O1 coordinates movement, posture and manipulation to autonomously perform household tasks like wiping tables, picking up trash and putting shoes away on request.
Related event: Light-O1: Humanoid Robot Brain Pretrained on Human Videos(2 posts)→
More from Embodied
- Tesla Optimus audits Chinese suppliers Tuopu, Sanhua, Joyson — with Optimus orders already placed — CyberRobooo · 2026-09-21
- Shuttered robotics company Eidon AI open-sources 9TB of egocentric video with IMU tracking — vanstriendaniel · 2026-09-21
- FANUC demos a cobot that reroutes around people using RL and NVIDIA cuRobo — lukas_m_ziegler · 2026-09-21
- NeurIPS 2026 EEG/EMG challenge: decode hand pose from wrist EMG with Meta Neural Band — agramfort · 2026-09-21
- Phone video in, navigable 3D room out: K3-powered splat pipeline runs in the browser — willeastcott · 2026-09-21
- World's first human vs. humanoid robot fight held, a 6ft robot steps into the ring — Admirable-Cell-2658 · 2026-09-21